Upload
jitenkg
View
217
Download
0
Embed Size (px)
Citation preview
7/23/2019 Loss ppt QC
1/157
$$: Swiss NSF, Nano Center Basel, EU (RTN), DARPA & ONR, ICORP-JST
Spin Qubits in Quantum Dots
Daniel Loss
Department of Physics
University of BaselSwitzerland
7/23/2019 Loss ppt QC
2/157
G. Burkard
B. CoishH.A. EngelV. GolovachD. Klauser
M. TrifD. DiVincenzoC. EguesO. Gywat
M. LeuenbergerJ. LevyF. MeierP. RecherE. Sukhorukov
S. TaruchaB. AltshulerD. AwschalomL. Glazman
L. Kouwenhoven
G. Abstreiter
L. SamuelsonL. VandersypenC. MarcusB. WesterveltC. Bruder
D. BulaevN. BonesteelM.S. ChoiK. Ensslin
A. ImamogluJ. LehmannA. MacDonaldD. Saraga
J. SchliemannC. SchnenbergerD. StepanenkoM. BorhaniR. Hanson
J. ElzermanF. Koppens
Special thanks to many Collaborators:
7/23/2019 Loss ppt QC
3/157
Outline
A. Spin qubits in quantum dots
1. Basics of quantum computing and quantum dots
2. Quantum gates: interaction based and measurement based entanglement
B. Spin decoherence in GaAs quantum dots
Spin orbit interaction and spin decay Alternative spin qubits: holes, graphene, nanotubes,... Nuclear spins and hyperfine induced decoherence
classical spin bath vs quantum and non-Markovian behavior
C. Nuclear spin order in 2DEGs (ultra-low temp.)
Kondo lattice in marginal Fermi liquid
7/23/2019 Loss ppt QC
4/157
Quantum Information
Classical digital computernetwork of Boolean logic gates, e.g. XOR
bits:
physical implementation:e.g. 2 voltage levels
gate: electronic circuit
Quantum computer
physical implementation:quantum 2-level-system:
quantum gate: unitary transformation(is reversible!)
1,0, =ba
1,10,22=++= ba
1,0
qubits
7/23/2019 Loss ppt QC
5/157
Quantum Computing (basics)
basic unit: = state of a quantum two-level systemqubit
"natural" candidate: electron spin
1) prepare N qubits (input)2) apply unitary transformation in 2N-dim. Hilbert space computation
3) measure result (output)
quantum computation:
"quantum" computation faster than "classicalfactoring algorithm (Shor 1994): exp N N2
database search (Grover 1996): N N1/2
7/23/2019 Loss ppt QC
6/157
Electron qubit: spin better than charge
natural choice for qubit: spin of electron
12 TT
echspin arg >>
1 ns
due to longer relaxation/decoherence* times
Awschalom et al., 97
Tarucha et al., 02Kouwenhoven et al., 03-07
Abstreiter et al., 04Zumbuhl & Kastner et al., '08
10ns -1s
mesoscopicsFujisawa et al. 03Marcus et al. 01
*)
theory: for single spin in GaAs dot (in principle)
GaAs
7/23/2019 Loss ppt QC
7/157
Spin qubits in quantum dots key elements
Initialization 1 electron, low T, high B0
Read-out convert spin to charge
then measure charge
ESR pulsed microwave magnetic field
SWAP exchange interaction
H0 ~ i zi
HJ ~ Jij (t) i j
HRF ~ Ai(t) cos(i t) xi
Coherence spin-orbit interaction
nuclear spins
EZ=
gBB
EZ=
gBB
J(t) J(t)
SL S
R
DL & DiVincenzo, PRA 1998
7/23/2019 Loss ppt QC
8/157
Electrical control and detection
Tunable # of electrons
Tunable tunnel barriers
Electrical contacts
A quantum dot as a tunable artificial atom
Confinement
Discrete # charges
Discrete orbitals
Delft group (L. Kouwenhoven & L. Vandersypen)
7/23/2019 Loss ppt QC
9/157
C. Marcus et al., PRL 2004
GaAs/AlGaAs Heterostruktur
2DEG 90 nm depth, ns = 2.9 x 1011 cm-2 Temp.: 100 mK
7/23/2019 Loss ppt QC
10/157
Double Dot in Carbon Nanotube (SW)
Schnenberger group (Grber et al., Phys. Rev. B 74, 075427 (2006))
7/23/2019 Loss ppt QC
11/157
Quantum Computing with Spin-QubitsDL & DiVincenzo, PRA 57 (1998)
H(t)=J(t) SLS
R
SL SR
1. Quantum gates based on exchange interaction:
Spin 1/2 of electron = qubit
7/23/2019 Loss ppt QC
12/157
Quantum Computing with Spin-QubitsDL & DiVincenzo, PRA 57 (1998)
SL SR H(t)=J(t) SLSR
CNOT (XOR) gate
s
dttHi
s duringJeTU
s
0,)(0
)('
=
=
x0
01
a b
7/23/2019 Loss ppt QC
13/157
Quantum Computing with Spin-QubitsDL & DiVincenzo, PRA 57 (1998)
H(t)=J(t) SLSRSL SR
CNOT (XOR) gate
2/12/122 121
SW
Si
SW
SiSi
XOR UeUeeUz
zz
=
:SWU
+ i
SW eU :2/1
switching time: 180 psPetta et al., Science, 2005
7/23/2019 Loss ppt QC
14/157
Quantum XOR gate with 'sqrt of swap'
,2/12/1)2/()2/( 121
SW
Si
SW
SiSi
XOR UeUeeUzzz +=
| | | |
| (|+i|) (i |+ |) |
i| (|-i|) (i |- |) -i|
i| | - | -i|
i| i | i| -i|
2/1
SWU
z
Sie 1
2/1
SWU
zSi
e2
2
zSi
e1
2
2
4/ie
2
4/ie
2
4/iie
2
4/iie
7/23/2019 Loss ppt QC
15/157
Quantum XOR gate with 'sqrt of swap'
2/1
SWU
| |+i|Square-root-of-swap:
Entanglement is crucial for quantum computing!
entangled state= entangler: product state
= |1
x |2
,2/12/1)2/()2/( 121
SW
Si
SW
SiSi
XOR UeUeeUzzz +=
7/23/2019 Loss ppt QC
16/157
Note: Control of exchange interaction J and switching timeneeds to be very precise (1:104)
experimental challenge
CNOT gate without interaction?
Yes: CNOT gates based on measurement:
Linear optics & single-photon detection conditional sign flip (non-deterministic) [1]Full Bell state analyzer & GHZ state deterministic quantum computing [2]
Partial Bell-state (parity) measurements deterministic quantum computing [3]
[1] E. Knill, R. Laflamme and G. J. Milburn, Nature 409, 46 (2001).
[2] D. Gottesman and I.L. Chuang, Nature 402, 390 (1999).[3] C.W.J. Beenakker et al., Phys. Rev. Lett. 93, 020501 (2004).
7/23/2019 Loss ppt QC
17/157
CNOT gate can be implemented with two parity gates
Beenakker, Kindermann & DiVincenzo, PRL 04
=
7/23/2019 Loss ppt QC
18/157
Deterministic entangler:
P
a
b
c
d
a,b: input armsc,d: output arms
baba
input statein arm a
input statein arm b (ancilla)
( ) ( ) ( ) ( )bababababbaa +++=++
Thus, we get:
Beenakker et al., 2004
Projective measurement: measurement of parity p projects input state into
either parallel output state (p=1) or antiparallel output state (p=0). If p=0, thenapply x(d) on output state get always same final output state in arms c and d.
1, =+ pifdcdc
dcdcdcdc pif +=+ ,0,
dcdcaa ++
7/23/2019 Loss ppt QC
19/157
| + || - | even parity Bell state: parallel spins
| + || - |
odd parity Bell state: antiparallel spins
Q: Does scheme exist for electron spins tomeasure parity of Bell states non-destructively?
Measurement-based quantum computing
with spin qubitsEngel & DL, Science 309, 586 (2005)
Advantage:parity measurement is digital (0 or 1) quantum gate is digital
7/23/2019 Loss ppt QC
20/157
Double Quantum Dot and QPC
Current IQPC depends on charge state1
odd parity: tunneling
even parity: no tunneling
IQPC
Quantumpoint contact
[1] J.M. Elzerman et al., Nature 430, 431 (2004)
IQPC
IQPC
I
I
L R
7/23/2019 Loss ppt QC
21/157
7/23/2019 Loss ppt QC
22/157
Tunneling transitions (coherent)
Single level picture
Many processes forsmall level spacing
But for>>Uonly transitions (a)
and (d) are relevant
|SL(a)
|SL(b)
|SR(c)
|T0L(d)
|T0L(e)
|T0L(f)
|T0
L
(g)
7/23/2019 Loss ppt QC
23/157
Phases due to different Zeeman interaction
during virtual occupation of state LR
|LR vs|LRcorrectable via one-qubit gatessuppression via large tdand fast read-out
Detuning from resonance
increases measurement time
Finite exchangeJ for LL and RR
additional dynamical phasecorrectable via one-qubit operations
Tunneling tStT and/orJLJR
|S and |T0 are distinguishable decoherence!
Imperfections
7/23/2019 Loss ppt QC
24/157
7/23/2019 Loss ppt QC
25/157
Universal Quantum Computing with Parity Gates
Two qubit dots (1 and L), paritymeasurement using reference dot (R)
(1, 1, 0) (0, 2, 0) (0, 0, 2)Transfer back adiabatically to (1,1,0)
1
L
R
P =
Parity gate for spin :(Engel & Loss, Science 05):
CNOT gate can be implemented with twoparity gates (Beenakker et al., PRL 04):
=
7/23/2019 Loss ppt QC
26/157
Perform parity gate (some details)
12
R2
QPC
time
1 2 3
p
7/23/2019 Loss ppt QC
27/157
a
t
R2
c
R1
QPC 1
QPC 2
time
1 2 3
p1
4 5 6 7 8 9 10
p2
H
H
H
H
m
11
Protocol for CNOT gate (1)
c
t
2121 tcctcUCNOT =
H
Engel and DL, 05
7/23/2019 Loss ppt QC
28/157
Protocol for CNOT gate (3): Conditional operations p1, p2, m represent the outcome of the measurements of steps 3, 7, and 10.
Detection of even parity (superposition of |00 and |11) is labeled as 0; odd parity as 1
In step 11, single-qubit gates listed on the rhs are applied to qubits c and t
I stands for identity (do nothing), X for X, Z for Z
p1 p2 m c t
0 0 0 I I0 0 1 I X
0 1 0 Z I
0 1 1 Z X
1 0 0 I X
1 0 1 I I
1 1 0 Z X
1 1 1 Z I
7/23/2019 Loss ppt QC
29/157
CNOT-Gate in Topological Quantum Computing via Parity(braiding of non-abelian anyons, Kitaev 03)
Zilberberg, Braunecker,and Loss, PRA 77, 012327 (2008)
Ising anyons, Bravyi 06
7/23/2019 Loss ppt QC
30/157
i
i
iBiji
ij
ij StBgSStJHrrrr
+= ))(()(
2-qubit gate'sqrt of swap' (s 100 ps)Petta et al., Science '05
1-qubit gatevia ESR for single spin (s 10 ns)Koppens et al., Nature '06, PRL '07;
via EDSR (T2 10 s), Science '07
Spin-Qubits in Dots are Scalable (?)
7/23/2019 Loss ppt QC
31/157
i
i
iBiji
ij
ij StBgSStJHrrrr
+= ))(()(
2-qubit gate'sqrt of swap' (s 100 ps)Petta et al., Science '05
1-qubit gatevia ESR for single spin (s 10 ns)Koppens et al., Nature '06, PRL '07;
via EDSR (T2 10 s), Science '07
Spin-Qubits in Dots are Scalable (?)
T2/
s~ 103 -105 *)
*) Coish & DL, PRB 75 (2007): s for single spin can be < 1ns (via exchange)
i.e. system is scalable
7/23/2019 Loss ppt QC
32/157
Spin-Qubits from Electrons
Many more choices for spin qubits:
SL S
Rsimplest spin-qubit:
spin-1/2 of 1 electron == 1,0
'exchange only qubits' DiVincenzo et al., 20003 electrons:
'singlet-triplet' qubits Levy 2002, Taylor et al., 2005
2 electrons: 'spin-cluster qubits' Meier, Levy & DL, `03
N coupled electrons: AF spin chains, ladders, clusters,...
molecular magnets Leuenberger & DL, 01; Affronte et al., 06Lehmann et al., 07; Trif et al., 08
== + 01,0 TTS
01,0 TS ==
7/23/2019 Loss ppt QC
33/157
Quantum Computing with Spin-Qubits
SL SR
Need to understand the dynamics and decoherencemechanisms for electron spins in quantum dots
CNOT (XOR) gate based on
entanglement such as
++ ii eore
i.e. phase coherence is crucial
7/23/2019 Loss ppt QC
34/157
Spin decoherence in GaAs quantum dots
1) Spin-orbit coupling (Dresselhaus & Rashba)
interaction between spin and charge fluctuations
Relaxation with rate 1/ T1
+ Decoherence with rate 1/ T2= decay of coherent superposition
Two important sources of spin decay in GaAs:
2) Hyperfine interaction between electron spin and nuclear spinsleads to non-exponential decay
G l i H il i
7/23/2019 Loss ppt QC
35/157
( )tgH B hSBS +=where h(t) is a fluctuating (internal) field with < h(t) >=0
( ) ( ) ( ) ( )[ ]
+= htEi
YYXX ZethhthhdtT 00Re1
1
( ) ( )
+= thhdtTT ZZ 0Re2
11
12
Relaxation (T1) and decoherence (T2) times in weak coupling:
General spin Hamiltonian:
[if < hi(t) hj(t) >~ij,i,j=(X,Y,Z)]
relaxation
contribution
dephasing
contribution
See e.g. Abragam
7/23/2019 Loss ppt QC
36/157
( )tgH B hSBS +=where h(t) is a fluctuating (internal) field with < h(t) >=0
( ) ( )
+= thhdtTT
ZZ 0Re2
11
12
For SOI linear in momentum:
General spin Hamiltonian:
relaxationcontribution
dephasingcontribution
Golovach, Khaetskii, DL, PRL 04
7/23/2019 Loss ppt QC
37/157
$$: Swiss NSF, Nano Center Basel, EU (RTN), DARPA & ONR, ICORP-JST
Spin Qubits in Quantum Dots
Part II
Daniel Loss
Department of PhysicsUniversity of Basel
Switzerland
7/23/2019 Loss ppt QC
38/157
All-electrical control and read-out achieved
Spin qubits in GaAs dots present status
1-qubit gate electron spin resonance
gate duration ~ 25 ns; observed 8 periods
2-qubit gate exchange interactiongate duration ~ 0.2 ns; observed 3 periods
Phasecoherence
T2* ~ 20 nsT
2
> 1 s
See also Hanson et al., Rev. Mod. Phys. 2007
Initialization 1-electron, low T, high B0
Read-out via spin-charge conversion
Energy
relaxation
T1~ 1 secduration ~ 100 s; 82-97% fidelity
duration ~ 5 T1; 99% fidelity ?
7/23/2019 Loss ppt QC
39/157
Outline
A. Spin qubits in quantum dots
1. Basics of quantum computing and quantum dots
2. Quantum gates: interaction based and measurement based entanglement
B. Spin decoherence in GaAs quantum dots
Spin orbit interaction and spin decay Alternative spin qubits: holes, graphene, nanotubes,... Nuclear spins and hyperfine induced decoherence
classical spin bath vs quantum and non-Markovian behavior
C. Nuclear spin order in 2DEGs (ultra-low temp.)
Kondo lattice in marginal Fermi liquid
7/23/2019 Loss ppt QC
40/157
Spin orbit interaction in GaAs quantum dots
interaction between spin and charge fluctuations SOI is weaker in quantum dots than in bulk since dot = 0
(Khaetskii & Nazarov, `00; Halperin et al., `01; Aleiner & Falko, `01)
e.g. spin-phonon: Khaetskii & Nazarov, PRB 64 (2001)Golovach, Khaetskii & Loss, PRL 93 (2004)
Fal'ko, Altshuler, Tsyplyatev, PRL (2005)Bulaev & Loss, PRL `05 (hole spin)
( ) ( ) )( 3pOppppH yyxxxyyxSO +++=
Dresselhaus SOIRashba SOI
7/23/2019 Loss ppt QC
41/157
Basics on Spin-Orbit Interaction
Relativistic (Einstein) correctionfrom Dirac equation:
=m
pVs
mcHso
rr
22
1
MeV12 2 mc
p
( )p
free electrons
eV1gE
k
( )k
semiconductor (Zinc-blende)
conduction band
heavy holes
light holes
bandstructureeffects
R hb i bit i t ti i 2D
7/23/2019 Loss ppt QC
42/157
Rashba spin-orbit interaction in 2D
spintronics
Rashba & Dresselhaus spin-orbit in 2DEG
7/23/2019 Loss ppt QC
43/157
E. Rashba, Nature Physics 2, 149 ('06)
Rashba & Dresselhaus spin orbit in 2DEG
7/23/2019 Loss ppt QC
44/157
Rashba & Dresselhaus Spin-Orbit in 2DEG
( )xyy
xR ppH =
( )yyx
xD ppH =
zE~Structure-inversion asymmetry tunable bygates: (Rashba, 1960)
Bulk-inversion asymmetry (Dresselhaus, 1955)
7/23/2019 Loss ppt QC
45/157
Rashba & Dresselhaus Spin-Orbit in 2DEG
( )xyy
xR ppH =
( )yyx
xD ppH =
zE~Structure-inversion asymmetry tunable bygates: (Rashba, 1960)
Bulk-inversion asymmetry (Dresselhaus, 1955)
=
is conserved spin (new basis) which makesspin of electrons independent of their momentum ( spin FET ?
tune Rashba term s.t.
yxDR ppHH m=+ ( )yx
( )yx
Spin-Orbit Interaction in GaAs Quantum Dots (2DEG):
7/23/2019 Loss ppt QC
46/157
( )xyyx
yyxxSO
pp
ppH
++=
p ( )
Dresselhaus S.-O.
Rashba S.-O.
( )tUHHHH phelSOZdot +++=
Model Hamiltonian:
( )rUm
pHdot += *
2
2
B = BZ gH 21
( ) ...= tU
phel any potential fluctuation, e.g., phonons
( )2
r
*
2
~m
h
piezoelectric & deformationacoustic
Electron-phonon interaction (2D):
7/23/2019 Loss ppt QC
47/157
Electron phonon interaction (2D):
( ) ( )( ) += +
j
jjjj
qjc
i
zphel bbiqeeqFU
, /2
q
qqqq
rq
h
zq,qq=
( )( ) ( )( )( )qqqjj
j eqeqqq
+=
2
2
piezo-electric interaction:
( )
( ) ( )
( )( )qqqjj
j eqeqq
+= 2
1
deform. potential interaction:
for GaAs:1,0 jj =q and
=
=otherwise,0
)(,14 cyclicxyzh
Parameter regime:
7/23/2019 Loss ppt QC
48/157
Parameter regime:
*, mSOSO h=
7/23/2019 Loss ppt QC
49/157
Parameter regime:
In this regime, we find that
12 2TT =
--- the spin-orbit interaction in GaAs quantum dots causes
a spin decay with the largest possible decoherence time,T2=2T1.
Note that, generally, T22T1, and, usually, one expectsT2
7/23/2019 Loss ppt QC
50/157
Parameter regime:
(typically: 100 nm, and SO1-10 m
2*2 mTkB h
7/23/2019 Loss ppt QC
51/157
( )tUHHHH phelSOZdot +++=( ) ( )[ ]tUStUHHHeeH phelphelZd
SS
+++= ,
~
[ ] SOZd HSHH =+ ,with S defined bywith px=im*[Hd,x], get [ ] == ddSO LiHiH ,
( ) ( ) ......
1
1 102
++=
+=
+= SSLi
L
L
LLi
LLS d
d
Z
d
d
Zd
no orbital B effect to O(Hso):
( ) ,10 = iPS nEnH nd =, =n
nn nnAAP
( )
( ) ( )[ ] 0,,0
== nnphelnnphel tUitUS
1st order in HSO
leading order is due to Zeeman term (no orbital):
7/23/2019 Loss ppt QC
52/157
( )( ) ,2
1
eff BB += tgH
B
( ) ( ),2 tt BB =
( ) ( ) ,/, 1 SOpheld tULt =
( )
,
1
11
=
= B
d
BZ
d L
PgiLL
PS
constanttindependenspin~
eff += HH
giving the effective Hamiltonian:
where
( ),0,',' += xy ( ) ,1*
h = m( )
( )
=
+=
2'
2'
yxy
yxxwith
Linear SO & Zeeman transverse fluctuations only:
7/23/2019 Loss ppt QC
53/157
( ) ( ) ''' aa +== teHetH tiHSOtiHSOZZ
,, ba == ZSO HH
( ) ( ) ( )[ ][ ]nanaaa +=+= tttHSO '''''
bbn=( ) ( ) ,''' aaannnana +=
first term commutes with HZ
( ) ( ),'''' tt aa =since rotation is around n.with ( ) ,0'' =na t
i.e. Zeeman interaction induces only
fluctuations transverse to b
( ))( tpaa=
go to rotating frame:
)('' tb
Now:
)('' tb
'''' aa RUU =+
7/23/2019 Loss ppt QC
54/157
1st order in Hso (time dependent perturbation theory):
( ) ( )[ ][ ]nan + tdttxdt
ddtctHdt
TT
SO
T
'')(000
SO linear : a ~ p ~ dx/dtnot a total derivative
for (He-ph)n, n>0
const. no orbital B field effect *)
Thus: if SO interaction linear in pwe get only transverse
fluctuations no pure dephasing T2=2T1
*) note: no orbital B field effect--only Zeeman, see alsoKhaetskii & Nazarov 00, Halperin et al., 01, Aleiner & Falko, 01
Derive Bloch equation for spin dynamics:
7/23/2019 Loss ppt QC
55/157
SSBS += Bg&
( ) ( ) ( )
=
0
2
22
02
dtetBBg
wJ iwt
jiB
ijij
h
spectral function
( ) ( ) ( ) ,Re wJwJwJ ijijij = ( ) ( ) ( )wJwJwI ijijij =
Im
relaxation:
dephasing:
decay: ,dr +=
( ) ( ) ( ) ( ) ( ) ( ), ++
+= qjpipqpqkkpqijpjpiippqqppqijr
ij IlIlJllJll
( ) ( )00 ++ = pjpipqqpijd
ij JllJll 0
c=/ s=100 ps
7/23/2019 Loss ppt QC
56/157
quantum well piezo deformation
( ) ( )( )
( )SOj
j
j
j
sz
j jc
zXX
sze
szFe
dsm
NzzJT
j 222
2
222
2
2sin
2
0
3
522
0
*
32
1
/cos
sin2
12Re1
22
+
+=
+
h
22
22
2
2
'
2
2
'
2
2
'
2
2
' 411112
+++
+
+
=
zyxyx lllll effective SO length
Bgz B=Golovach, Khaetski, Loss, PRL 93 (2004)
Relaxation rate:Bose function super-Ohmic: ~z
3
7/23/2019 Loss ppt QC
57/157
smsssms /1037.3,/107.4 3323
1 =
quantum well piezo deformation
( ) ( )( )
( )SOj
j
j
j
sz
j jc
zXX
sze
szFe
dsm
NzzJT
j 222
2
222
2
2sin
2
0
3
522
0
*
32
1
/cos
sin2
12Re1
22
+
+=
+
h
13,/16.0,sin)1cos3(23
,2sin2,cossin23,7,
2
14
21
14
2
1
14
221
14
2
001,
2
,3
,2,1
=
===
mChh
hheVjj
22
22
2
2
'
2
2
'
2
2
'
2
2
' 411112
+++
+
+
=
zyxyx lllll effective SO length
speed of sound
Bgz B=
Spin relaxation rate 1/T1 for GaAs Quantum Dot
7/23/2019 Loss ppt QC
58/157
Spin relaxation rate 1/T1 for GaAs Quantum Dot
( ) 22 2sin2
0
122
1
sin)(1
jB sBgk
B edBgT
++
22 BB 2
)( ph pHSO
power-5 law for B< 3T (GaAs)
Golovach et al., PRL 93 (2004)
Numerical value of T1 for GaAs parameters (13!):
7/23/2019 Loss ppt QC
59/157
)()0020.00077.0(04.043.0 TBg =Hanson ea PRL 91 (196802) 200329.0=gor with linear fit:
Theory: T8Bfor,s7501 =T
( )
===
e
cSO
m
mssshmd
067.0,mkg103.5,scm1035.3,scm1073.4
,mC16.0,eV7.6,1.13,0,m9,nm5,nm32,meV1.1,,,,,,,,,,,
3355
2
*
321140
*
0
hh
Zumbuhl ea PRL 89 (276803) 2003
TsT [email protected] =Experiment:Elzerman et al.,
Nature 430, 431 (2004)
22/SO
s11005501 =T due to uncertainties in gfactorms7.21=T for SO=17m [Huibers ea PRL 83, 5090 (1999)]
Current record: T1 > 1 s (B 1 T) in GaAsS A h K M L I R d D Z b hl
7/23/2019 Loss ppt QC
60/157
S. Amasha, K. MacLean, I. Radu, D. Zumbuhl,
M. Kastner, M. Hanson, A. Gossard, Phys. Rev. Lett. 100, 46803 (2008).
data in good agreement with theory
Golovach, Khaetskii, DL, PRL 93 (`04)
Rashba & Dresselhaus SOI Effectswell understood (?)
Spin relaxation rate 1/T1 for GaAs Quantum Dot
7/23/2019 Loss ppt QC
61/157
( ) 22 2sin2
0
5
1
sin)(1
jB sBgk
B edBgT
7/23/2019 Loss ppt QC
62/157
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.40
1
2
3
T
1
[ms]
S-T splitting [meV]
Theory: Golovach, Khaetskii, Loss, PRB 2007
Experiment on double-dots, Meunier et al., PRL 2007
phonon wavelength
matches dot size
p 1
7/23/2019 Loss ppt QC
63/157
Brute force nuclear polarization with large B-field:
Note: spin phonon rate is non-monotonic function of B-field
Chesi & DL, cm0804.3332
1/T1,2 depends strongly on B-field direction (magic angles):
7/23/2019 Loss ppt QC
64/157
( )( )0,2/
,,1
11 ===
T
f
T
( ) ( )( )[ ]
2sinsin2cos11
,, 22222
+++=f
43,2, === 1Texact!
Special case:
Rashba and Dresselhaus interfere! *)
*) Schliemann, Egues, DL, PRL `03
Golovach, Khaetskii, LossPRL 93, 016601 (2004)
ellipsoid
1/T1,2 depends strongly on B-field direction(magic angles):
7/23/2019 Loss ppt QC
65/157
( )( )0,2/
,,1
11 ===
T
f
T
( ) ( )( )[ ]
2sinsin2cos11
,, 22222
+++=f
43,2, === 1Texact!
Special case:
Rashba and Dresselhaus interfere! *)
*) Schliemann, Egues, DL, PRL `03
Golovach, Khaetskii, LossPRL 93, 016601 (2004)
ellipsoid
Relaxation of spin in GaAs quantum dots dominated by
7/23/2019 Loss ppt QC
66/157
Relaxation of spin in GaAs quantum dots dominated by
spin-orbit & phonons with ultra-long relaxation times T1:
T1 ~ 1s for B ~ 1TAmasha et al.,Phys. Rev. Lett. (2008)
Relaxation of spin in GaAs quantum dots dominated by
7/23/2019 Loss ppt QC
67/157
e a at o o sp Ga s qua tu dots do ated by
spin-orbit & phonons with ultra-long relaxation times T1:
But measured spin decoherence times are muchshorter: T2~1- 10 s Petta et al. '05; Koppens et al. '06/'07
T1 ~ 1s for B ~ 1T
From Rashba- SOI we expect T2 = 2T1 Golovach et al., PRL '04
Thus, spin decoherence in GaAs must be dominated byother effects hyperfine interaction with nuclear spins
Burkard, DL, DiVincenzo, PRB 99
Amasha et al.,Phys. Rev. Lett. (2008)
Hyperfine interaction:
7/23/2019 Loss ppt QC
68/157
Major source of dynamics/decoherenceQuantum dots Si:P donors
Molecular magnetsNV centers in diamond
Ono and Tarucha, PRL (2004),Petta et al., Science (2005),
Koppens et al., Nature (2006)E Abe et al., PRB (2004)
Childress et al., Science (2006),Hanson et al., PRL (2006)
Ardavan et al., PRL (2007)Barbara et al., Nature (2008)
...All are potential candidates for quantum information processing applications
7/23/2019 Loss ppt QC
69/157
1. Avoid nuclear spin problem: use holes or othermaterials such as C, Si, Ge,...
2. Use GaAs (still best material) and dealwith nuclear spins
Strategies:
7/23/2019 Loss ppt QC
70/157
1. Avoid nuclear spin problem: use holes or othermaterials such as C, Si, Ge,
2. Use GaAs (still best material) and dealwith nuclear spins
Strategies:
Heavy Hole Spins in Quantum Dots
fl d HH LH i i d l li d HH
7/23/2019 Loss ppt QC
71/157
Heiss et al., Phys. Rev. B 76,2413062 (2007);
[and: Gerardot et al., Nature451, 44108 (2008)]
flat dot HH-LH mixing suppressed long-lived HH
Bulaev & Loss, PRL 95, 076805 (2005)
Abstreiter/Finley group 07: T1 ~ 200 s
self-assembled InGaAs dots
Hyperfine effect on holes??
Hyperfine Interactions with Nuclear-Spins
7/23/2019 Loss ppt QC
72/157
Hyperfine Interactions with Nuclear Spins
Fermi contact hyperfine interaction
Anisotropic hyperfine interaction
Coupling of orbital angular momentum
kkjS
k
ISrh k
rrr
= )(38
4
01
)/1())((3
4 30
2
kk
kkkkjS
k
rdrISInSnh
k +=
rrrrrr
)/1(43
03
kk
kkjS
k
rdr
ILh
k +
=
rr
Band structure of a GaAs QW
7/23/2019 Loss ppt QC
73/157
Band structure of a GaAs QW
In the quasi-2D limit, a gapdevelops between the HH andLH sub-bands.
The HH-LH degeneracy islifted.
For GaAs:
eV
eVeVE
LH
SO
g
1.0
3.05.1
=
==
(QW height of5nm)
s-type
p-type
Hyperfine Interactions with Nuclear-Spins
7/23/2019 Loss ppt QC
74/157
yp p
Fermi contact hyperfine interaction
Anisotropic hyperfine interaction
Coupling of orbital angular momentum
kkjS
k
ISrh k
rrr
= )(38
4
01
)/1())((3
4 30
2
kk
kkkkjS
k
rdrISInSnh
k +=
rrrrrr
)/1(43
03
kk
kkjS
k
rdr
ILh
k +
=
rr
Electronsh1 isotropic
Holesh2,3 Ising-like
h2,3 ~ 0.2 h1
Fischer, Coish, Bulaev, Loss, arXiv:0807.0368
Carbon-based Nanostructures
7/23/2019 Loss ppt QC
75/157
Folklore: 12C is light atom and thus weak spin-orbit interactionis it true?
)( 3 yyxx ppvH +=
2D Dirac-Hamiltonian:
Spin Qubits in Graphene
7/23/2019 Loss ppt QC
76/157
Trauzettel, Bulaev, Loss & Burkard, Nature Physics 3, 192 (2007)
Advantages: weak SOI and
(almost) no nuclear spins;& long-range interaction
Challenge: armchair boundariesto lift orbital degeneracy
Carbon Nanotube Quantum Dots
7/23/2019 Loss ppt QC
77/157
Nygard, Cobden, Lindelof, Nature 408, 342-6 (2000).Jarillo-Herrero, et al., Nature 429, 389 (2004).Mason, et al., Science 303, 655 (2004).Graber, et al., Phys. Rev. B 74, 075427 (2006).
Carbon Nanotube:
7/23/2019 Loss ppt QC
78/157
Carbon Nanotube:
Dirac-like bandstructure withsmall gap: semiconducting NT:
Gate-controlled confinment:
Spin orbit interaction: zparaiperp
SOI ScheSiH 132 2.).( ++= +
Ando, J. Phys.Soc. Jpn., 2005;
Huertas-Hernando et al., PRB 2006
Spectrum of Nanotube Quantum-Dot in B-fieldD. Bulaev, B. Trauzettel, and D. Loss, PRB 77, 235301 (2008)
7/23/2019 Loss ppt QC
79/157
Spin-Orbit Interaction leads to zero-field splitting
Spectrum of Nanotube Quantum-Dot in B-fieldD. Bulaev, B. Trauzettel, and D. Loss, PRB 77, 235301 (2008)
7/23/2019 Loss ppt QC
80/157
Spin-Orbit Interaction leads to zero-field splitting
Spectrum experimentally
confirmed, see Kuemmeth, Ilani,Ralph, McEuen, Nature 452 (2008)
Spectrum of Nanotube Quantum-Dot in B-fieldD. Bulaev, B. Trauzettel, and D. Loss, PRB 77, 235301 (2008)
7/23/2019 Loss ppt QC
81/157
Spectrum experimentally
confirmed, see Kuemmeth, Ilani,Ralph, McEuen, Nature 452 (2008)
Spin-Orbit Interaction leads to zero-field splitting all-electrical control of spin possible!
Spin Relaxation and Spin Decoherence Rates in Nanotubesdue to Electron-Phonon and Spin Orbit Interactions
7/23/2019 Loss ppt QC
82/157
Extreme variations withB-field: spin relaxation rate
can be varied by a factor 108
!
1/f phonon noise
due to quadratic bending modes ~ q2
in 1D
Bulaev et al., PRB 77, 235301 (2008)
7/23/2019 Loss ppt QC
83/157
1. Avoid nuclear spin problem: use holes or othermaterials such as C, Si, Ge,
2. Use GaAs (still best material) and dealwith nuclear spins
Strategies:
Spin-bath dynamics
7/23/2019 Loss ppt QC
84/157
Fully-polarized bath exact dynamics
Thermal (completely random) bath
Spin-echo decay
Semiclassical dynamics
Khaetskii, Loss, Glazman, PRL (2002), ...: power-law or inverse-log decay
Khaetskii, Loss, and Glazman, PRL (2002), Merkulov, Efros, andRosen, PRB (2002), Coish and Loss (2004), ...: Gaussian decay
Erlingsson and Nazarov, PRB (2004), Yuzbashyan et al., J. Phys. A.(2005), Al-Hassanieh et al., PRL (2006), Chen, Bergman, and Balents,PRB (2007), ...: power-law or inverse-log decay
de Sousa and Das Sarma, PRB (2003), Witzel and Das Sarma, PRL(2007), Yao, Liu, and Sham, PRL (2007), ...: super-exponential decay
BUT: Non-exponential decay inconsistent with Markovianquantum error-correction models!!
zvv
Hyperfine Interaction in Single Quantum Dot
7/23/2019 Loss ppt QC
85/157
x
y
zB( )rr
z
( ) nKsHdd 1001014
2/12
nuclear spindipole-dipole interaction
ddB
i
ii HSBgISAH ++=
electron Zeeman energy
2)( ii rAA
v
hyperfine interactionis non-uniform:
516 10,10 = NsN
A
Khaetskii, DL, Glazman, 02; Coish & DL, 04-07; Eto 04; Sham 06; Altshuler 06;Balents 07 ...
Separation of the Hyperfine Hamiltonian
7/23/2019 Loss ppt QC
86/157
Hamiltonian: VHhSBSgH zB +=+= 0vv
= iiiIAh
vv
Note: nuclear field
... ... ... ...
zzB ShBgH )(0 +=
( )++ += ShShV
2
1
longitudinal component
flip-flop terms
Separation:
yx ihhh =V V
is a quantum operator
Nuclear spins provide hyperfine field hwith
( ) fl i b l i
7/23/2019 Loss ppt QC
87/157
(quantum) fluctuations seen by electron spin:
Sv
hv
Nuclear spins provide hyperfine field hwith
( t ) fl t ti b l t i
7/23/2019 Loss ppt QC
88/157
(quantum) fluctuations seen by electron spin:
Sv
hv
Nuclear spins provide hyperfine field hwith
( t ) fl t ti b l t i
7/23/2019 Loss ppt QC
89/157
(quantum) fluctuations seen by electron spin:
With mean =0 and (quantum) varianceh:
1
2
1
2
)10(5/
= ===
== nsmTNAIAhh nucl
N
k
kknucl
r
what state?
Svh
v
Initial conditions for nuclear spinsCoish &DL, PRB 70, 195340 (2004)
7/23/2019 Loss ppt QC
90/157
[ ] === knnz
z
kkzI nhnIAnhnn ,)0(
)3(
Pure state of hz-eigenstate:
Initial conditions for nuclear spinsCoish &DL, PRB 70, 195340 (2004)
7/23/2019 Loss ppt QC
91/157
( )
( )
=
=
+==
N
NNNI
k
i
k
N
kIIII
NNffNN
fef k
1)0(
1,)0(
)2(
1
)1(
[ ] === knnz
z
kkzI nhnIAnhnn ,)0(
)3(
Superposition (1) or mixture (2) of hz-eigenstates:
Pure state of hz-eigenstate:
Initial conditions for nuclear spinsCoish &DL, PRB 70, 195340 (2004)
7/23/2019 Loss ppt QC
92/157
[ ] === knnz
z
kkzI nhnIAnhnn ,)0(
)3(
( )
( )
=
=
+==
N
NNNI
k
i
k
N
kIIII
NNffNN
fef k
1)0(
1,)0(
)2(
1
)1(
Superposition (1) or mixture (2) of hz-eigenstates:
Pure state of hz-eigenstate: Pure state of hz-eigenstate:
nuclear spin bath behaves classically !
Gaussian decay and narrowing
Zeroth order in flip-flop terms: )]0([Tr )( thBgi zBeSS +
7/23/2019 Loss ppt QC
93/157
Zeroth order in flip-flop terms:
Gaussian decay
)]0([Tr )(0 I
thBgiIteSS +++ =
,22 2
0
.)(cttti
measno
t eeSS
++
zh
h
B
nsp
N
Atc 5
1
22
= h
GaAs, N=105
C h t i l t t i l t
(tunneling)
E. Laird et al., PRL '06: free induction decay
7/23/2019 Loss ppt QC
94/157
Theory (full lines): Coish & DL,PRB 72, 125337 (2005)
Coherent singlet- tripletoscillations due to hyperfinemixing with Gaussian distribution reasonable agreement
between theory and experiment
Koppens et al., PRL 2007: driven Rabi oscillations
7/23/2019 Loss ppt QC
95/157
Gaussian Hyperfine field :
(Klauser, Coish & DL)
Gaussian decay and narrowing
Zeroth order in flip-flop terms: )]0([Tr )( IthBgiIzBeSS +++ =
7/23/2019 Loss ppt QC
96/157
Zeroth order in flip flop terms:
Prepare nuclear spin system with a measurement of the Overhauser field
B. Coish and D. Loss (PRB 2004), G. Giedke et al. (PRB 2006)D. Klauser et al. (PRB, 2006), D. Stepanenko et al. (PRL 2006)
Gaussian decay
)]0([Tr0 IIt
eSS ++ =
,22 2
0
.)(cttti
measno
t eeSS
++
zh
h
B
Precession
[ ] nnzBtimeas
t hBgeSS += ++
,0
.)(
B
h
nsp
N
Atc 5
1
22
= h
GaAs, N=105
Narrowing of nuclear spins in double dots with ESR
Klauser, Coish & DL, PRB 73, 205302 (2006)
ESR: oscillating exchange J(t) J +j cos(t) leads to Rabi oscillations:
7/23/2019 Loss ppt QC
97/157
z
nB hBg +
ESR: oscillating exchange J(t)=J0+j cos(t) leads to Rabi oscillations:
+=
=
ESR at frequency = measures eigenvalue nuclear spins projected into corresponding eigenstate |n>
z
nB hBg + z
nh
If quantum measurement is ideal, then Gaussian superpositioncollapses to a single Lorentzian (ESR linewidth):
j/2
Quantum control of
7/23/2019 Loss ppt QC
98/157
0
0.2
0.4
0.6
0.8
1
1.2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 20
0.2
0.4
0.6
0.8
1
1.2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
0
0.2
0.4
0.6
0.8
1
1.2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 20
0.1
0.2
0.3
0.4
0.5
0 20 40 60 80 100
b)a)
c)
M = 51, = 1
M= 100, = 22
M= 50, = 0
(x x0)/0
(x x0)/0
(x x0)/0
M
d)
(x) (x)
(x)P
Quantum control ofnuclear spin bath
through measurementof single electron spin
Klauser, Coish & DL, PRB 73, 205302 (2006)
7/23/2019 Loss ppt QC
99/157
[ ]
===
knnz
z
kkzI
nhnIAnhnn ,)0()3(
Consider now nuclear spins in pure state of hz-eigenstates:
hv
Narrowed initial nuclear spin state h=0 at t=0
7/23/2019 Loss ppt QC
100/157
hSv .constIS
k
k=+rr
Khaetskii, DL, Glazman, PRL 02 & PRB 03
Coish &DL, PRB 70, 195340 (2004)
t>0: quantum dynamics ... ... ... ...V V
changes hyperfine field in time by 1/N spin precesses influctuating hyperfine field spin dephases (power law decay)
flip-flops
back action of Son hBz
hv
IS
Narrowed initial nuclear spin state h=0 at t=0
7/23/2019 Loss ppt QC
101/157
hSv
Dynamics (flip-flops): ... ... ... ...V V
.constISk
k=+rr
2/3
/
2
2
)/()(4
)0()(
NAt
e
pIAbN
AStS
NitA
zz
+
Time scale is N/A = 1s (GaAs) and decay is bounded
power law decay
back action of Son h
E.g.
Bz
Khaetskii, DL, Glazman, PRL 02 & PRB 03, Coish &DL, PRB 04
Quantum vs. Classical Dynamics
7/23/2019 Loss ppt QC
102/157
N
A
IKKSBSH
N
kk
z
==+= = ;; 1vvv
Special case: Uniform hyperfine coupling =kA
)()()(
)()()(
tstktk
tstkBts
vv&vv
vv&v
=
+=
Classical: Time-dependentmean-field (TDMF)
Exact solutioncf. Yuzbashyan and Altshuler,
J. Phys. A (2005)
Quantum:
Exact solutioncf. Coish, Yuzbashyan, Altshuler, Loss
J. Appl. Phys. 101, 081715 (2007);
[ ][ ]KHiK
SHiSv&v
v
&
v
,
,
==
t
t
K
Sv
v
Initial Conditions
Classical Quantum
7/23/2019 Loss ppt QC
103/157
( )
( )sssss
kk
s
k
cos,sinsin,cossin)0(
cos,0,sin)0(
=
=
v
v
( )==
+=
=
m
k
K
mK
iK
k
siss
ks
mKdKKe
e
ky
s
,,2
sin
2
cos
)0(
)(
Classical Quantum
)0(sv
)0(kv
B
Sv
Kv
B
classical vectors spin coherent states('minimum uncertainty')
( )0=B
Classical vs. Quantum Dynamics
7/23/2019 Loss ppt QC
104/157
( )
( ) N
AT
tsS
tsStd
TtC
xtx
xtxTt
t
==+
=
+
;1.0
;21
)(22
( )
( )10=B
TDMF:
quantum:
)(tsx
txS
Time-dependent mean-field (TDMF) = exact quantum solution1)( =tC
)(tsS xtx for times t> AN
c
2=1)(
7/23/2019 Loss ppt QC
105/157
Time
21
~
b
A
N
A
Nc ~
A
N
A
b~
???~2T 2Tt>>
tx
Fully polarized nuclei: exactly solvable case
BSIS AH
Khaetskii, DL, Glazman, PRL `02 & PRB `03
7/23/2019 Loss ppt QC
106/157
The initial nuclear spin configuration is fully polarized. With the initial wavefunction 0 we construct the exact wave function of the system for t > 0 :
Normalization condition is: , and we assume that
From the Schrdinger equation we obtain:
paramagnon
entangled
BSIS += Bii
i gAH
(1)
7/23/2019 Loss ppt QC
107/157
Laplace transform of (1) gives:
++
+
+=
++=+==
k
k
k
iu
iuuD
iut
uDi
uDtiu
2/A4/)2A(
A
4
1
4
)2A()(
,2/A4/)2A(
)0(A)(2)(
)0()(
kz
2
kz
kz
k
])2
)(A1(lnN2i-A[2
2/A
A 20
k
2
k =+
N
zidz
ik
Introducing 4/)2A( z ++= iiu , using (t=0)=1, k (t=0)=0,and replacing the sum over k by an integral
self-energy
(1)
7/23/2019 Loss ppt QC
108/157
where A = kA k; l=1,, N set of N+1 coupled differential eqs. (N>>N)
Spin correlator: 2/))(1()()(2
000 tStStC zz ==
Laplace transform of (1) gives:
here
note: sums k replaced by integrals over rk3 (valid for t < N2/A),
with x,y (Gaussians) integrated out non-analyticity
integration contour and singularities
7/23/2019 Loss ppt QC
109/157
The singularities are: two branch points (=0, 0= i A02(0)/ 2N), and firstorder poles which lie on the imaginary axis (one pole for z > 0, two poles for
z < 0). For the contribution from the branch cut (decaying part) we obtain:
)0(20 =here and .)0()(),(2
00
2
000 == zzz
g g
1) Large Zeeman field |z| >> A :
Asymptotic behavior ( = t/(N/A) >>1) determined by nuclei at dot center:
7/23/2019 Loss ppt QC
110/157
i.e. power law decay and bounded by 1/N and (A/z)2
The decay law depends on the magnetic field strength . However, the
characteristic time scale for the onset of the non-exponential decay is the same
for all cases and given by (A/N )-1 (microseconds in GaAs dot).
~ 1/ N
2) No external magnetic field (z= 0):Asymptotic behavior ( >> 1) determined by weakest coupled nuclear spins:
2/3ln/1~ non-universal
z
Bi
ii SBgISAH +=
vv
General case: below full polarization
7/23/2019 Loss ppt QC
111/157
x
y
zB
( )rr
z
Coish & Loss, 04-07; Eto 04; Sham 06; Altshuler 06; Balents 07,...
Generalized Master Equation
Generalized Master Equation (exact): [ ]OHLO ,=
Bill Coish &DL, PRB 70, 195340 (2004)
Liouville op.
7/23/2019 Loss ppt QC
112/157
tdtttitiLt
t
SSSS =
0
0 )()()()( &
)0()( IViQLt
IS LLeiTrt =
[ ] )0()()( 1)2(0)2( SSS siiLss
++=
Ge e a ed aste quat o (e act)
L++=+= )()()0(1
)()4()2(
ssLiQLsLiTrsSSIVIS
)0(11
)0(1
)0(1
00
IVVIIVIIVI LiQLs
QLiQLs
LTrLiQLs
LiTrLiQLs
LiTr ++
+
=+
Lowest-order Born approximation:
+
=i
i
S
st
iS dsset
)()( 21
Laplace transform, expand self-energy in the transverse coupling :VL
)()( tTrt IS =
[ ]
[ ]OVOL
OHOL
V ,
,00
=
=
( )OTrQO II )0(1 =
)0()0()0( IS =
nnI =)0(
[ ] nhnhnnzz
=
Initial conditions and perturbative regime
Initial State: )0()0()0( IS =1
El i[ ]=
nnzznhnh ,
7/23/2019 Loss ppt QC
113/157
j
z
jI mnInnn == ,)0(
zzyyxxSS SSSI 0002
1)0( +++=
Nuclear spins:
Electron spin:
=i
z
iiz
nnzz
IAh
Initial conditions and perturbative regime
Initial State: )0()0()0( IS =SSSI
1)0(El t i
[ ]=nnzznhnh ,
7/23/2019 Loss ppt QC
114/157
)(
)()( 0
sis
sNSsS
zz
zz
z ++=
)()( 0
siis
SsS
n ++
++ +=
longitudinal and transverse electron spin dynamics decouple to all orders:
j
z
jI mnInnn == ,)0(
zzyyxxSS SSSI 0002
1)0( +++=
Nuclear spins:
Electron spin:
=i
z
iiz IAh
[ ]nnzn
hb+=
Perturbative regime for all electron spin dynamics given by: IAb >>
( )1
12
0
2/1
0
2
7/23/2019 Loss ppt QC
115/157
0iLs +
[ ]
[ ]
[ ])()()(
)()()(
)()()(
)2(
)2(
)2(
sIcsIciNs
isIisIiNcs
isIisIiNcs
nn
nn
++++
+
++
+=
++=
++=
1,ln
2
4
1)(
1
0
2
== md
ixs
xxdx
m
d
Ais
A
NsI
k k
k
mm
=
m
l
rr
02
1exp)0()(
In d Dimensions:
),1:2/1(
)()()(
,)1()1(
+
===
=
+=
fcfcI
mFmPmF
mmIIc
m
I
d=m=2: [ ] isisssI = )log()log()( mbranch cuts
Dependence on the envelope wave function
1
Donor Impurity:
( )02/exp)( lrr
1exp)0()(
=m
rr
Envelope wave function:
7/23/2019 Loss ppt QC
116/157
0 10 20 30 40 50
t [2 --h/A0]
-0.5
0
0.5
Re[I
+(t)/I
0]
2D Quantum Dot:
[ ]sANt 1/2 h
( ) 2//exp)( 20lrr
)0(
)(
+
+
R
tR
.2
exp)0()(0
=l
r
,2,
1
)1(/
> m
d
ettR NiAt
m
d
X
,11,ln
)1(2
=>>
m
d
t
ttRX
in d dimensions.
Decaying fraction of the spin:
zXSStR XtXX ,,)( 0 +==
2
2
)(4 pIAbN
A
+=
Dot:
Si:P
Spin dynamics is `non-Markovian (for t~N/A)
Born-Markov Approximation not valid:
7/23/2019 Loss ppt QC
117/157
( )
t
ti
t
t
t
tti
ttSeSStttdiSiS nn +
+++++
++ ==
~
0
)2()( ,)(e~&
( )
( ) !0)~(Im
)~(Re~
)2(
)2(
=+==
+==
++
++
n
n
is
is
)(0
tRSeS t
t ++
+ +=
From the exact GME:
i.e. vanishing decay rate!
Markov approximation(in the rotating frame)
non-Markovianremainder term
0,1
)0( *
2
=+ zBBgNp
R
Stationary Limit & Decaying Fraction
p
Longitudinal spin: Stationary limit within the Born approximation (I=1/2):
7/23/2019 Loss ppt QC
118/157
+
+===
12)(lim
1lim
0
0
0
pS
ssSdtST
Sz
zs
T
tzTz
Np21
zS
Decaying fraction for b=0 is of order independent ofdimensionality or specific form of the electron envelope wavefunction.
This result is confirmed by exact diagonalization ofsmall systems, e.g., Ntot=15:
0 5 10 15 20t (2N/A)
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
t
)2
1,0(
122
==== IbNp
N
n
Summary of electron spin dynamics
zXSStR XtXX ,,)( 0 +== 2*2
)(4)(
pIABgNAtR
zB
X +=
7/23/2019 Loss ppt QC
119/157
=
m
l
rr
02
1exp)0()( in d dimensions, and for a wave function:
+= 1
* pIABg
A
zB
Exact when p=1,Born approximation valid in
the weakly perturbative regime.
7/23/2019 Loss ppt QC
120/157
Time
21
~
b
A
N
A
Nc ~
A
N
A
b~ ???~
2
T2
Tt>>
tx
Free-induction decay: Big picturePower law:
Khaetskii, Loss, Glazman, PRL (2002)Coish and Loss, PRB (2004)
tx
7/23/2019 Loss ppt QC
121/157
Time
21
~
b
A
N
Initial quadratic (Gaussian?) decay:Yao, Liu, Sham, PRB 06, PRL 07, NJP 07
2)/(2)/(1 tt etx
A
Nc ~
A
N
A
b~ ???~
2
T2
Tt>>
t
Free-induction decay: Big picturePower law:
Khaetskii, Loss, Glazman, PRL (2002)Coish and Loss, PRB (2004)
tx
7/23/2019 Loss ppt QC
122/157
Time
???
21
~
b
A
N
Initial quadratic (Gaussian?) decay:Yao, Liu, Sham, PRB 06, PRL 07, NJP 07
2)/(2)/(1 tt etx
A
Nc ~
A
N
A
b~ ???~2T 2Tt>>
t
Free-induction decay: Big picturePower law:
Khaetskii, Loss, Glazman, PRL (2002)Coish and Loss, PRB (2004)
tx
7/23/2019 Loss ppt QC
123/157
Time
???
21
~
b
A
N
Initial quadratic (Gaussian?) decay:Yao, Liu, Sham, PRB 06, PRL 07, NJP 07
2)/(2)/(1 tt etx
A
Nc ~
A
N
A
b~ ???~2T 2Tt>>
Long-time power law:Deng and Hu, PRB (2006)
2/1 tx t
t
Consider now times t >> N/ACoish, Fischer, and Loss, Phys. Rev. B 77, 125329 (2008)
Coherence of an electron spin can exhibit exponential
7/23/2019 Loss ppt QC
124/157
p pdecay for large magnetic field and narrowed initial
bath-state Hyperfine Hamiltonian
2nd order effective Hamiltonian (Schrieffer-Wolff trafo)
Reproduces 4th-order rates when expanded to 2nd order in flip-flopsonly
)(2
1)( ++ ++++= ShShIbShbH
k
z
kk
zz
hf
+
+++=
+
k
z
kk
z
lk
lkz
lkz IbSIIhb
AAhbH
2
1
Details of the Transformation
Write the hyperfine Hamiltonian as
( )++ +++ ShShVSHVHH z 1
7/23/2019 Loss ppt QC
125/157
Schrieffer-Wolff transformation
This can be rewritten as
( )+=+=+= ShShVSHVHH zhf2
,, 000
VL
SVSHeHeH S
hf
S
0
0
1],,[
2
1=+=
( ) DSXSSHSSHH z ++=+= ++
diagonal off-diagonal
(with respect to product eigenbasis of )z
kI
Hyperfine-Mediated Nuclear-Field Fluctuations Heff
)()( 3VODSXH zeff +++=
7/23/2019 Loss ppt QC
126/157
Effective (fluctuating) field leads topure dephasing of electron spin
Fixed frequency(narrowed initial state)
(small for large b)
= 10],[ TSH z
Beff ~ +X
2/Tt
tx eS
zhb +
z
k
k
kID
bIIAA
X lklk
lk
1
2
1
+
Generalized Master Equation (GME)
Initial conditions:
nnnn === )0()0()0()0(
7/23/2019 Loss ppt QC
127/157
Nakajima-Zwanzig GME:
Expand in powers of
nnnn nIIs === ,)0(,)0()0()0(
'0
)'('t
t
tn
t
SttdtiSiSdt
d+++
=
bXSV z 1~=
...)()()()4()2(
++= ttt 1
7/23/2019 Loss ppt QC
128/157
Markovian regime:
+ 0 )(Re, tSex tt
)()(
~ )(tet
ti n
= +
2Tc
7/23/2019 Loss ppt QC
129/157
Markovian regime:
+ 0 )(,tt
)()(
~ )(
tet
ti n
= +
2Tc
7/23/2019 Loss ppt QC
130/157
Markovian regime:
+ 0 )(,tt
)()(
~ )(
tet
ti n
= +
2Tc
7/23/2019 Loss ppt QC
131/157
Homonuclear system:
Continuum limit:
=
02
)0()(Re1
XtXdteT
ti titiXeetX
=)(
tAAi
l
lk
klkeAA
bXtX
)(22
2
1)0()(
tAAi
lklkeAAdldk
b
)(22
002
1 ANc /
)0()( XtX
t
Distribution of coupling constants
k spins
N spinsd=1
(e.g. nanotube, nanowire)
7/23/2019 Loss ppt QC
132/157
d=2
d=3
volume occupied by one spin
(e.g. lateral quantum dot)
(e.g. Self-assembled dot, defect)
Decoherence Rate:
Homonuclear System
7/23/2019 Loss ppt QC
133/157
Coupling toone nucleus
Geometricalfactor
N
A
b
A
q
df
II
T
22
2 3
)1(1
+=
4~I 1
7/23/2019 Loss ppt QC
134/157
Markovian
Non-Ma
rkovian
1D quantum dot (e.g. nanotube, nanowire)
donor impurity
2D quantum dot(e.g. lateral gated dot)
2T
=
q
Ba
rr exp)0()(
Heteronuclear system
(neglecting inter-species flip-flops*) =
i
iiT
2
2
1
7/23/2019 Loss ppt QC
135/157
Isotopic abundance
IIn =9/2 IGa,As =3/2
InxGa1-xAs, x=0.05
*) if difference in nuclear
Zeeman energy > A/N
explains strong isotopedependence seen in transport
measurements (Ono+Tarucha,04; Yacoby 07)
Free-induction decay due to HyperfinePower law:
Khaetskii, Loss, Glazman, PRL (2002)Coish and Loss, PRB (2004)
2Long-time power law:
tx
7/23/2019 Loss ppt QC
136/157
Time
???
21
~
b
A
N
Initial quadratic (Gaussian?) decay:Yao, Liu, Sham, PRB 06, PRL 07, NJP 07
2)/(2)/(1 tt etx
A
Nc ~
A
N
A
b~ ???~2T 2Tt>>
g pDeng and Hu, PRB (2006)
2/1 tx t
Power law:Khaetskii, Loss, Glazman, PRL (2002)Coish and Loss, PRB (2004)
2
Long-time power law:tx
Free-induction decay due to Hyperfine
7/23/2019 Loss ppt QC
137/157
Time
21
~
b
A
N
Initial quadratic decay:Yao, Liu, Sham, PRB 06
2)/(1 tx t
A
Nc ~
A
N
A
b~ NbT
2
2
2 ~ 2Tt>>
g pDeng and Hu, PRB (2006)
2/1 tx tExponential decayCoish et al., PRB (2008)
2/Tt
t ex
Summary of Part A&B
spin qubits and entanglement: all-electrical control spins in GaAs dots: long relaxation times (secs!)
7/23/2019 Loss ppt QC
138/157
spins in GaAs dots: long relaxation times (secs!)
due to SOI & phonons but short decoherence due to hyperfine interaction alternative systems: holes, graphene, nanotubes,...
power law decay for times up to N/A exponential decoherence with strong isotope
dependence ( for t > N/A, and large B)
1. Dynamical polarization
C. Polarization of nuclear spins
7/23/2019 Loss ppt QC
139/157
ESR & optical pumping:
7/23/2019 Loss ppt QC
140/157
A ~90eV in GaAs
A/EF
~ 10-2
7/23/2019 Loss ppt QC
141/157
Schrieffer-Wolff transformationintegrate out electron degrees of freedom
RKKY interactionbetween nuclear spins
(overrules direct dipolarinteraction)
electron spin susceptibility
A ~90eV in GaAs
A/EF
~ 10-2
7/23/2019 Loss ppt QC
142/157
Look at stability by considering spin wave excitations
electron spin susceptibilityfor noninteracting Fermi gas
2D
What is required for magnetic order?
Assume a ferromagnetic ground state(ensured by Weiss mean field theory)
7/23/2019 Loss ppt QC
143/157
Look at stability by considering spin wave excitationsspin wave excitation
spectrum
2D2D
What is required for magnetic order?
Assume a ferromagnetic ground state(ensured by Weiss mean field theory)
7/23/2019 Loss ppt QC
144/157
Look at stability by considering spin wave excitationsspin wave excitation
spectrum
continuum of zero energy excitations
nuclear order unstable no magnetization!
2D
But if...
Ferromagnetic order would be stable up to some T > 0if the spin wave dispersion was linear at |q| 0
Required: Non analytic behavior
7/23/2019 Loss ppt QC
145/157
Required: Non-analytic behavior
Many-body corrections from electron-electroninteraction beyond standard Fermi-liquid theory?
Non-analyticities in Spin Susceptibility in 2DEG
Consider screened Coulomb U and 2nd order pert. theory in U:
)q(
,0 A typical diagram
7/23/2019 Loss ppt QC
146/157
Chubukov, Maslov, PRB 68 ('03)Aleiner, Efetov PRB 74 ('06)
,q
7/23/2019 Loss ppt QC
147/157
Possibility of spiral order
But it has the wrong sign!
Negative spin wave dispersion.
7/23/2019 Loss ppt QC
148/157
But is this generally true? Based on perturbation theory &screened Coulomb interaction
Shekhter & Finkelstein, PRB`06;
Braunecker, Simon, Loss, PRB08
Chesi, Zak, DL, in preparation
NO! Renormalization important in 2D!
Possibility of spiral order
Renormalization
Backscattering strongly renormalized by Cooper channelcontribution (P 0 & presence of Fermi sea).
7/23/2019 Loss ppt QC
149/157
is possible
Leads to renormalization of scattering amplitude: () (,q)Saraga, Altshuler, Loss, Westervelt, PRB 05;Chesi, Zak, Loss, 08
Renormalize lowest order diagrams
Chesi, Zak, DL, in preparation
7/23/2019 Loss ppt QC
150/157
Chubukov, Maslov, PRB 68, 155113 (2003)
Renormalization
Include Cooper channel in S(q) (Kohn-Luttinger instability w.r.t. q):
7/23/2019 Loss ppt QC
151/157
renormalized scatteringamplitude
Saraga, Altshuler,Loss, Westervelt
PRB 05
dominating Cooper channel: = (: angle between incident electrons)
Since
if
for small |q|
Braunecker, Simon, Loss, PRB08
Ferromagnetic order is possible
Nuclear Ferromagnetstable
7/23/2019 Loss ppt QC
152/157
Also a possibility
7/23/2019 Loss ppt QC
153/157
Possibly spiral order
Critical Temperature
Nuclear Ferromagnetstable
shape obtained also
7/23/2019 Loss ppt QC
154/157
Finite magnetization at low temperatures -value of Critical Temperature?
shape obtained also
by Local Field Factorcalculation for
long-range Coulombinteractions
Critical Temperature
Nuclear magnetization at finite temperature
7/23/2019 Loss ppt QC
155/157
valid atwith the Curie Temperature
Finite magnetization at finite temperature in 2D!
Estimate for GaAs 2DEG: Tc ~ 25 K 1 mK, for rs ~ 0.8 - 8
7/23/2019 Loss ppt QC
156/157
ERR
OR
:
syn
ta
xerr
or
OFFENDIN
G
CO
MMAND
:--n
ostrin
STA
CK
:
/Ti
tl
e
()
/Subject
(D
:2
00807070
85744+
02
00
)
/M
odD
ate
()
/K
eyw
or
ds
Cr
eator
Ver
si
on
0.9.5)
/Cr
eator
(D
:2
00807070
85744+
02
00
)
/Cr
eati
onD
at
e
(Ammini
str
at
ore)
/A
uth
or
-m
ark-
7/23/2019 Loss ppt QC
157/157
gv
al--